MBE Advance Access originally published online on December 28, 2006
Molecular Biology and Evolution 2007 24(3):827-835; doi:10.1093/molbev/msl211
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Research Articles |
One Billion Years of bZIP Transcription Factor Evolution: Conservation and Change in Dimerization and DNA-Binding Site Specificity
,
,
* Faculty of Life Sciences, University of Manchester, United Kingdom
Department of Ecology and Evolution, University of Lausanne, Switzerland
Swiss Institute of Bioinformatics, Lausanne, Switzerland
Bioinformatics Division, Institute for Evolution and Biodiversity, School of Biological Sciences, University of Muenster, Germany
E-mail: david.robertson{at}manchester.ac.uk.
 |
Abstract |
|---|
 TOP Abstract IntroductionMaterials and MethodsResults and DiscussionConclusionsSupplementary MaterialAcknowledgementsReferences |
|---|
The genomic era has revealed that the large repertoire of observed animal phenotypes is dependent on changes in the expression patterns of a finite number of genes, which are mediated by a plethora of transcription factors (TFs) with distinct specificities. The dimerization of TFs can also increase the complexity of a genetic regulatory network manifold, by combining a small number of monomers into dimers with distinct functions. Therefore, studying the evolution of these dimerizing TFs is vital for understanding how complexity increased during animal evolution. We focus on the second largest family of dimerizing TFs, the basic-region leucine zipper (bZIP), and infer when it expanded and how bZIP DNA-binding and dimerization functions evolved during the major phases of animal evolution. Specifically, we classify the metazoan bZIPs into 19 families and confirm the ancient nature of at least 13 of these families, predating the split of the cnidaria. We observe fixation of a core dimerization network in the last common ancestor of protostomes–deuterostomes. This was followed by an expansion of the number of proteins in the network, but no major dimerization changes in interaction partners, during the emergence of vertebrates. In conclusion, the bZIPs are an excellent model with which to understand how DNA binding and protein interactions of TFs evolved during animal evolution.
Key Words: bZIP • dimerization • transcription factor • molecular evolution • network • bilateria
|
Introduction |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResults and DiscussionConclusionsSupplementary MaterialAcknowledgementsReferences |
|---|
As organisms increase in complexity, they appear to require more transcription factors (TFs). Indeed, it is argued that complexity correlates with an increase in both the absolute number of TF genes as well as in the proportion of TFs in a genome (Levine and Tjian 2003
; van Nimwegen 2003; Ranea et al. 2005). This has been proposed to allow the decoupling of complexity and gene number, increased complexity being achieved by more sophisticated control of available genes as well as by alternative splicing. It has been proposed that the apparent explosion of bilaterian animal forms in the Cambrian was not just the result of extensive gene duplication events, but was specifically due to the expansion and change of function of a few key gene families that form the kernels of genetic regulatory networks (Miyata and Suga 2001; Kusserow et al. 2005; Davidson and Erwin 2006).
One major way of creating a large repertoire of regulatory responses is by the formation of TF dimers with distinct DNA binding and dimerizing properties, allowing the emergence of novel control circuits (Klemm et al. 1998). Such a fundamental importance of dimerization in creating complex dynamic behaviors within a cell has been confirmed by mathematical modeling (Smolen et al. 2000). In order to understand how animals evolved, it is important to delineate the mechanisms of evolution at the DNA-binding and protein interaction levels of these dimerizing TFs.
Major superfamilies of dimerizing networks, such as the basic helix-loop-helix (bHLH), basic-region leucine zipper (bZIP), and nuclear receptors, regulate a very wide range of processes, such as cell cycle, reproduction, development, homeostasis, metabolism, and programed cell death (Gu et al. 2000; Massari and Murre 2000; Luscher 2001; Wagner 2001; Robinson-Rechavi et al. 2003; Gronemeyer et al. 2004). Our previous work on the bHLH dimerization network (Amoutzias et al. 2004) revealed its hub-based topology and proposed a model of single-gene duplications, domain rearrangements, and point mutations for its emergence and fixation in the early metazoa, followed by subsequent gene duplication (including large-scale duplication events) in the vertebrate lineage. An obvious question that arises is whether or not the same general properties and evolutionary mechanisms apply to other heterodimerizing TF networks.
Here, we focus on the bZIP proteins, which are dimerizing TFs found in all eukaryotes. They are involved in a wide range of functions, such as development, metabolism, circadian rhythm, learning, memory, and response to stress and radiation, thereby acting as environmental biosensors as well as controllers of development (Wagner 2001; Deppmann et al. 2006). There is a wealth of data on DNA binding and dimerization in bZIPs. These provide an excellent model to understand how these functions evolved during the major phases of animal evolution: the emergence of multicellular animals around 1 billion years ago, the emergence of bilaterian animals around 650 MYA, and the emergence of vertebrates around 550 MYA.
bZIPs take their name from their highly conserved bZIP domain and comprise a basic region (BR) and a leucine zipper (LZ) (Hurst 1994). The BR makes contact with the DNA element (hexamer) and is highly conserved. The metazoan bZIPs can recognize 6 major categories of consensus DNA-binding sites, the TPA responsive element (TRE), AMP responsive element (CRE), CAAT box, AF recognition element (MARE), CRE-like, and PAR binding sites, whereas fungal bZIPs can recognize the TRE-, CRE-, and YAP-binding sites (Deppmann et al. 2006). The LZ is a short amphipathic coiled-coil protein domain responsible for recognition and dimerization specificity and is less conserved than the BR (Fassler et al. 2002). There is a wealth of data on protein–DNA binding and also a very extensive data set of protein–protein interactions among the bZIP proteins. The rules of dimerization specificity among all the members of this family have been deciphered at an unprecedented level of detail in the last few years (Fassler et al. 2002; Vinson et al. 2002; Newman and Keating 2003; Fong et al. 2004; Grigoryan and Keating 2006; Vinson et al. 2006).
The bZIPs from humans, fruit flies, and ascidians have been identified and classified (Fassler et al. 2002; Vinson et al. 2002; Yamada et al. 2003). More recently, we have discussed the topology and properties of the bZIP dimerization network compared with other dimerizing TF families and have shown that its architecture is not random, but linked to redox control and blocking of DNA binding (Amoutzias et al. 2006). In addition, a recent review discussed in detail the DNA binding and interactions of bZIPs in several distant taxa (Deppmann et al. 2006). Although these studies used data from several species, they did not study the evolution of the bZIPs in detail; rather, they focused on understanding the rules of dimerization specificity and the topology and function of the dimerization network. Open questions include the following: 1) How did the bZIPs expand during the major phases of animal evolution? 2) How did the DNA-binding and the dimerization networks evolve during these phases? 3) What insight can bZIPs give us into animal evolution?
In this study we identify, using hidden Markov models (HMMs), detectable bZIP homologues in several chordate, arthropod, cnidarian, and fungal genomes and classify these bZIPs using phylogenetic analysis. For the first time, we include phylogenetic data from bony fishes and the cnidarian lineage, enabling insights into events at the origin of vertebrates, at the origin of bilaterian animals, at the origin of metazoa, and finally at the split of fungi–metazoa. We integrate our phylogenetic profile analysis with domain architecture and protein–protein interaction data in order to increase confidence in the phylogeny and elucidate in detail the mechanisms of evolution and degree of conservation in the bZIP dimerization network. Our results indicate that genome duplication, consistent with 2R (Wolfe 2001), has had a major role in the emergence of the bZIP network. In addition, gene duplication and major dimerization changes in interaction partners, "rewiring," occurred during the emergence of metazoa and of bilaterian animals. This rewiring was followed by fixation of a core network architecture. Finally, we discuss the evolution of DNA-binding specificity and summarize the history of these ancient molecules.
|
Materials and Methods |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResults and DiscussionConclusionsSupplementary MaterialAcknowledgementsReferences |
|---|
Custom HMMs
HMMs (Eddy 1996
) were generated that could detect all of the known metazoan and fungal bZIP proteins in genome searches. In a preliminary analysis, we compared the speed, sensitivity, and specificity of custom-made bZIP HMMs with an analysis that detected bZIPs using Blast (Vinson et al. 2002) and observed that custom-made HMMs were producing fast and equivalent results with fewer false positives. Thus, the HMM, specific for the DNA-binding and LZ domains, is superior to a Blast search. In order to increase the sensitivity of the HMMs, we generated bZIP-specific HMMs corresponding to their phylogenetic classification. In this way, accurate detection and classification of sequences from available genomes was achieved.
The bZIP HMMs were generated as follows. A training set of 214 sequences that were annotated as bZIPs were downloaded from the TRANSFAC database, public release 4 (Matys et al. 2003). These sequences were used to create multiple alignments using T-COFFEE (Notredame et al. 2000), based on their classification in TRANSFAC. The alignments were manually edited using PFAAT (Johnson et al. 2003) where necessary. HMMs were generated using the HMMER package, version 2.3.2 (available from http://hmmer.janelia.org). These HMMs were used to retrieve bZIP sequences from SWISSPROT version 40 (Boeckmann et al. 2003). From this data set, new HMMs were generated for each of the 11 previously identified bZIP families and for 1 generic model including all the bZIPs.
bZIP Designation
The criteria used for bilaterian bZIP designation were 1) presence of a BR, defined by Vinson et al. (2002), and 2) presence of a LZ. The presence of a BR is mandatory because LZ-like sequences can apparently emerge de novo (Brendel and Karlin 1989). The program 2ZIP (Bornberg-Bauer et al. 1998) was used to identify LZs that were located at the C terminus of the highly conserved BR. The LZ region had to be within the 6 heptad repeats located C terminally to the BR and have a coiled coil (as predicted by the 2ZIP program) of at least 3 heptads, 21 amino acids (aa), within the first and sixth heptad. These criteria were relaxed for the cnidarian sequences because this part of the analysis was based on DNA trace files.
Protein Sequence Retrieval
The custom HMMs were used to scan the following genomes using the HMMPFAM software with a cutoff value of 1 x 10–5: Homo sapiens (human), Takifugu rubripes (puffer fish), Danio rerio (zebra fish), Ciona intestinalis (ascidian), Drosophila melanogaster (fruit fly), Apis mellifera (honey bee), Anopheles gambiae (mosquito), Schizosaccharomyces pombe (archaeascomycete), Yarrowia lipolytica, Debaryomyces hansenii, Kluyveromyces lactis, Candida glabrata, and Saccharomyces cerevisiae (hemiascomycetes). Because there are no finished cnidarian or poriferan genomes available to date, a keyword search was performed for any bZIP proteins of these taxa in the National Center for Biotechnology Information (NCBI) protein database. The retrieved sequences were inspected to determine whether they complied with the criteria outlined for designation as a bZIP. Cnidarian, Hydra magnipapillata (hydra) and Nematostella vectensis (sea anemone), sequences were also obtained from NCBI's Genome Trace archive using TBlastN (http://www.ncbi.nih.gov). The retrieved sequences were translated in all 6 frames and then were scanned with our custom HMM models. Identical H. magnipapillata and N. vectensis sequences were excluded from the analysis.
Evolutionary Analysis
The identified bZIP sequences for each family were aligned with T-COFFEE (Notredame et al. 2000) and manually edited where necessary using PFAAT (Johnson et al. 2003). Phylogenetic trees were inferred by Neighbor-Joining (PROTDIST and NEIGHBOR) with the PHYLIP package (available from http://evolution.genetics.washington.edu/phylip.html) and by maximum likelihood with PHYML (Guindon and Gascuel 2003), using the Jones-Taylor-Thornton (JTT) model of amino acid replacements. The alignments are too short for meaningful bootstrap analysis. Tree visualizations were obtained with TreeDyn (available from http://www.treedyn.org).
Characterization of Domain Architecture
The domain architecture of bZIP protein sequences was determined, using annotation from the PRODOM database (Bru et al. 2005). For each sequence, a Blast search against the PRODOM sequence database was performed. Hits with an e value greater than 1 x 10–4 or with less than 60% conserved residues were rejected. This sequence identity threshold has been shown to be sufficient to identify related domains longer than 30 aa in length (Brenner et al. 1998). The annotated domains were not permitted to overlap. Finally, the results were inspected manually. If alternative annotations were possible, the annotation that contained a bZIP domain was used.
Network Construction
The bZIP dimerization network was obtained from a published protein array experiment (Newman and Keating 2003). In this study, the coiled coils of 49 human bZIP genes were checked for the presence of an interaction with any of the other bZIP coiled coils, and each protein was used both as a surface bait and a probe. Thus, each heterodimerization event is represented twice in the matrix and is associated with 2 Z scores. We considered an interaction as valid if its Z score was greater than 2.5 in both instances.
|
Results and Discussion |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResults and DiscussionConclusionsSupplementary MaterialAcknowledgementsReferences |
|---|
bZIP Classification
The evolutionary relationships between the identified bZIP sequences were inferred by constructing a phylogenetic tree based on the bZIP domain, specifically the BR and LZ domains (figs. 1 and 2). The relatively short length of these domains (
60 aa) hinders inference of phylogenetic relationships, particularly of the deeper divergences (ancient nodes). Nonetheless, the inferred relationships are consistent with previous phylogenetic analyses (Vinson et al. 2002; Yamada et al. 2003), and for the majority of clusters, orthology between monophyletic arthropod and chordate clades is apparent as expected (fig. 2). Clusters were also consistent between the Neighbor-Joining and maximum likelihood phylogenetic analyses (figs. 1 and 2). To increase our confidence in these clusters, we determined the domain architecture of all the bilaterian bZIP proteins, apart from those obtained with TBlastN from the DNA trace files of cnidarian genomes. These domain architectures were generally similar among the vertebrate members of each of the clusters (fig. 2), again confirming the reliability of the phylogenetic clusters.
|
|
We classified the bZIPs based on the phylogeny of their BR–LZ domains. bZIP families were defined by 1) the presence of vertebrate–arthropod orthologues, 2) well-defined clades, and 3) conserved common domain arrangement. Nineteen bZIP families can be identified (figs. 1 and 2). The CHOP protein was excluded due to the absence of a typical BR. The families inferred by the Neighbor-Joining analysis (fig. 2) were confirmed by the maximum likelihood analysis (fig. 1). Human bZIP proteins had previously been classified into 12 families based on sequence similarity and dimerization properties of the BR–LZ domain (Vinson et al. 2002
) or into 16 families based on the sequence similarity of the coiled coil (Newman and Keating 2003). These previously identified families are consistent with the 19 families identified here. Our new groupings are as follows: 1) the split of the OASIS family into OASIS and OASISb, 2) the split of the CNC family into NFE2 and BACH, and 3) the split of the C/EBP family into C/EBP and C/EBP-g. These designations are consistent with the phylogenetic analysis of Yamada et al. (2003) in which humans, C. intestinalis, and fly sequences were included. The BACH family does not contain any arthropod representative, but includes C. intestinalis, indicating that a common ancestor existed before the vertebrate split. BATF is the only family that does not have an arthropod or C. intestinalis orthologue; nevertheless, it forms a distinct phylogenetic cluster (figs. 1 and 2). This could be due to the emergence of this family in vertebrates or to gene losses in the other lineages, as has been observed elsewhere (Bertrand et al. 2004; Shiu and Li 2004; Popovici et al. 2005; Danchin et al. 2006).
Evolution during the Emergence of Vertebrates
Our comparison of the phylogenetic relationships of bZIP from multiple genomes not only gives insights into the ancient origins of the bZIP superfamily, but also permits the identification of more recent periods of evolutionary change. Specifically, each bZIP family (apart from BATF) is represented by at least a single bZIP in the last common ancestor of vertebrates and arthropods, a relationship, which has been largely maintained in arthropods. Indeed, all families, apart from one, are represented by a single bZIP protein in arthropods (fig. 2). Interestingly, in the last common ancestor of tetrapods and bony fish, the 19 subfamilies contained 1 (e.g., XBP1) to 4 (e.g., FOS) proteins, bringing the total number of bZIPs in the ancestral vertebrate genome to at least 35 (based on human/ray-finned fish orthology; see fig. 2). The inclusion of genomic data from ray-finned fishes and from C. intestinalis clearly shows that these additional bZIPs are the product of duplications at the origin of vertebrates, consistent with the hypothesis of 2 closely successive rounds of whole-genome duplication (the 2R hypothesis; Ohno 1970; Dehal and Boore 2005). If there were 18 ancestral families with at least 1 member, then (after the 2 rounds of whole-genome duplication and without gene loss) there should be 72 genes in total (4 x 18). However, it is known that few genes are retained after a large-scale duplication (Wolfe 2001). The fish/human orthologies indicate that at least 35 of these 72 genes were retained. Additionally, 4–5 genes seem to have duplicated in the tetrapod lineage, whereas for the other genes the picture is not clear. Thus, it appears that bZIPs are relatively resilient to the gene losses expected after genome duplication. Others have also noted that genes encoding TFs have a significant tendency to be retained after such a duplication, as compared with genes in other functional categories (Blomme et al. 2006).
An interesting observation is that the paralogues of each bZIP family share not all but very similar interaction partners (Newman and Keating 2003). This, coupled with the fact that most of these proteins predate the human/fish divergence, shows that the interfamily interactions and the overall topology of the network, as seen in humans, was already formed in the ancestor of bony vertebrates. Major features of the network topology were most probably formed before the genome duplication events that occurred in the vertebrate lineage. Therefore, whole-genome duplication did not significantly change the topology of the network. Rather, it added new paralogues that mostly retained their ancestral dimerization preferences, with new interactions being formed in a few cases. These vertebrate lineage duplications also did not have any major impact on the evolution of new specificities for recognition by DNA-binding motifs (supplementary fig. 1, Supplementary Material online).
A Conserved Core Dimerization Network Emerged before the Protostome–Deuterostome Split
How old is the architecture of the bZIP dimerization network? Seventeen of the 19 families have arthropod orthologues (the exceptions being BATF and BACH), indicating that the proteins in the network were already present in the last common ancestor of arthropods and vertebrates. Several of the interfamily interactions, such as the JUN–FOS, JUN–ATF3, and CNC–S-MAF, have been found in Drosophila from yeast 2-hybrid data (Giot et al. 2003), confirming the ancient character of this dimerization network. Thus, it appears that there is a highly conserved core network that was formed in the last common ancestor of arthropods and vertebrates, around 600 MYA. A significant level of network conservation has also been observed for the bHLH dimerization network, where the overall architecture is shared between humans and flies (Amoutzias et al. 2004).
This conclusion concerning the high degree of conservation of the basic layout of the bZIP dimerization network and of the biophysical properties of the bZIP's coiled coil over evolutionary time does not preclude other layers of complexity related to the functional role of bZIPs. For example, other types of factors, potentially include changes to the 1) TFs that regulate bZIPs, 2) downstream target genes of bZIPs, 3) posttranslational modifications, 4) emergence or loss of various interaction surfaces outside of the highly conserved bZIP domain, and/or 5) colocalization or coexpression with potential interaction partners. Such factors will potentially result in differences between bZIPs in ancestral and divergent lineages.
bZIP Family Duplication and Dimerization Rewiring at the Origin of Bilateria
The recently available genomic data from cnidaria combined with phylogenetic analysis, as well as functional data (DNA-binding specificities and dimerization), provide insights into the evolution of the network during the emergence of the metazoa, in the period 1,000 and 650 MYA (fig. 3). Specifically, the phylogenetic analysis (figs. 1 and 2) reveals ancient duplication events that shaped the topology of the bZIP network at the origin of metazoa. At least 10 families group in pairs (C/EBP–C/EBP-g, E4BP4–PAR, L-Maf–S-Maf, OASIS–OASISb, and NFE2–BACH), with each pair forming a distinct clade. Possibly, this pairing is due to another ancient whole-genome duplication, this time in the Precambrian period (discussed below). The biological significance of this pairing is further supported by the available DNA-binding data for at least 4 of the pairs (C/EBP–C/EBP-g, E4BP4–PAR, L-Maf–S-Maf, and NFE2–BACH). More specifically, the members of each pair have very similar DNA-binding domains and also recognize the same consensus DNA-binding sites with the highest affinity (supplementary fig. 1, Supplementary Material online; Haas et al. 1995; Oyake et al. 1996; Moll et al. 2002). Note, they can also identify other DNA-binding sites, but with lower affinity. For 3 pairs (C/EBP–C/EBP-g, NFE2–BACH, and OASIS–OASISb), both families of a given pair retain a large number of common interfamily interactions (supplementary fig. 2, Supplementary Material online), whereas each pair has also evolved some new interactions. In contrast, for each of the E4BP4–PAR and L-Maf–S-Maf pairs of families, the duplicates either lost a significant number of common interactions or evolved new interactions. Each of the 5 pairs of duplicate families retained their DNA-binding properties but diverged in the LZ domain after the duplication event, thus making new interactions or losing old ones. This resulted in an overall increase in protein numbers and the rewiring of the dimerization network, especially due to the duplication of the E4BP4–PAR and L-Maf–S-Maf pairs of families.
|
So, when did these ancient duplications occur? Our phylogenetic analysis indicates that 13 of the 18 ancient families (with vertebrate and invertebrate homologues) have orthologues also in cnidaria, a very close sister group of bilaterian animals (fig. 1). For the C/EBP–C/EBP-g, E4BP4–PAR, L-Maf–S-Maf, and NFE2–BACH pairs of families, either only 1 of the 2 pairs has cnidarian orthologues or the cnidarian sequences fall basal to the pair. This striking phylogenetic pattern suggests that, for at least the 3 pairs of families mentioned above (C/EBP–C/EBP-g, E4BP4–PAR, and L-Maf–S-Maf), duplication occurred after the divergence of cnidaria and before the vertebrate–arthropod split, placing these duplication events in the Cambrian period. The NFE2–BACH pair possibly duplicated later, at the emergence of chordates. More specifically, from the 13 ancestral families with orthologs in cnidaria, 3 ancestral families duplicated into 6, raising the total number to 16. The other ancestral family (ATF3) was either lost in the cnidarian lineage or was generated at the emergence of the bilateria. The OASIS–OASISb pair of families apparently duplicated before the split of the cnidarian lineage as both families cluster with cnidarian orthologues.
An alternative (and we propose less likely) scenario is that these duplication events occurred even earlier than the split of cnidaria, and the phylogenetic relationships observed (fig. 1) are misleading as a result of 1) gene losses in the cnidarian lineage, 2) insufficient coverage in the genomic trace files of H. magnipapillata and N. vectensis, and/or 3) insufficient detection of bZIPs from the trace files, due to splicing events in the middle of the bZIP domain. Kusserow et al. (2005) have shown that 11 of the 12 bilaterian Wnt signal transduction families are also found in N. vectensis, indicating that this sea anemone's genomic inventory was very similar to that of bilateria. Therefore, our findings do not support extensive gene losses of signaling families in sea anemones. Regarding insufficient coverage from the genomic trace files, given the fact that we found an average of 37 (many of them redundant) DNA reads (with identified bZIP domains) per family from N. vectensis, this seems an unlikely explanation. As for the possibility of splicing events in the middle of the bZIP domain, we checked the exon structure of the human genes (from ENSEMBL), for the missing families, and assuming a conserved exon structure, only the BACH family could possibly have escaped detection. Therefore, the scenario of gene duplications of 3 families at the emergence of bilateria is the most parsimonious explanation. Due to the extent and timing of this duplication event (which also enabled network rewiring), coupled with the fact that bZIPs control developmental processes, we propose that bZIPs are 1 of few key gene superfamilies that contributed to the explosion of animal forms in the Cambrian (Miyata and Suga 2001; Davidson and Erwin 2006).
The inclusion of cnidarian and fungal sequences (fig. 1) shows that at least 13 families date back to the common ancestor of cnidaria–bilateria and that 12 of these families are metazoan specific. Most probably, these 12 families emerged at the origin of multicellular animals, around 950 MYA. Miyata and Suga (2001) have shown that there was an extensive gene duplication and domain shuffling that generated the various signaling families at the origin of multicellular animals. Probably, the same event generated the 12 bZIP families that we find in cnidaria and that are metazoan specific. The imminent sequencing of the genomes of metazoan basal sponges and the unicellular sister group of multicellular animals, the choanoflagellates, will more accurately permit the timing of these major duplication events and permit an understanding, at the transcriptional level, of what happened at the emergence of multicellular life.
The protein interaction network that we have used here (supplementary fig. 2, Supplementary Material online) to infer ancestral networks (fig. 3) is derived from dimerization data from a protein array experiment in humans (Newman and Keating 2003) and so is an estimate of both 1) the true human bZIP network and 2) bZIP network of other vertebrates. When more complete experimental data are available for more organisms, it will be interesting to revisit this inference of ancestral network structures.
|
Conclusions |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResults and DiscussionConclusionsSupplementary MaterialAcknowledgementsReferences |
|---|
We have investigated the evolution of the dimerization network of the metazoan bZIP TF family, which is central to the emergence of transcriptional complexity. Sequence-based phylogenies, comparison of conserved patterns in DNA–protein–binding interfaces, domain arrangements, and protein–protein interactions have been combined to permit a new insight into the evolution of DNA binding and dimerization of TFs (summarized in fig. 3). bZIPs are very old proteins that evolved some of their DNA-binding specificities before the divergence of the metazoa and fungi. Later, at the origin of multicellular animals around 950 MYA, they probably underwent extended gene duplications that allowed them to evolve new DNA-binding specificities as well as complex dimerization networks. The recognition of new DNA surfaces, when coupled with heterodimerization, must have had a tremendous impact at the organismal level because it increased the complexity and inventory of possible regulatory responses manifold. Three to 5 gene duplication events, possibly at the origin of bilaterian animals, created new families. These retained their DNA-binding specificity but evolved new interactions, thus further increasing the complexity of the network. Seventeen of the 19 bZIP families were already present in the last common ancestor of arthropods and vertebrates, and a core dimerization network was fixed by that time. Later, at the emergence of vertebrates, 2 whole-genome duplication events that occurred in close succession increased the paralogues of each family, but these new paralogues have not undergone significant changes in their DNA-binding and dimerization specificities. Most of the changes must have happened outside of the bZIP domain, thus facilitating new interactions with signal transduction and coactivator–corepressor proteins. The majority of these ancient duplicates are retained in humans. In conclusion, the bZIPs are an excellent model family with which to understand how animal complexity evolved over a billion years at the gene regulatory level.
|
Supplementary Material |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResults and DiscussionConclusionsSupplementary MaterialAcknowledgementsReferences |
|---|
Supplementary figures 1 and 2 are available at Molecular Biology and Evolution online (http://www.mbe.oxfordjournals.org/).
|
Acknowledgements |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResults and DiscussionConclusionsSupplementary MaterialAcknowledgementsReferences |
|---|
G.D.A. was the recipient of a CASE studentship from the Engineering and Physical Sciences Research Council (EPSRC) and AstraZeneca and was also supported by an EPSRC platform grant (GR/R80810/01) to S.G.O. and others. A.V. was supported by a Marie Curie Training Site Fellowship to the University of Manchester. A.V. and E.B.-B. also acknowledge support from Deutsche Forschungsgemeinschaft (project grant BO-2544/2-1). Work on protein interactions in D.L.R.'s and S.G.O.'s laboratories is supported by research grants from the Biotechnology and Biological Sciences Research Council. G.D.A. gratefully acknowledges support from Dimitris and Vasiliki Amoutzias.
|
Footnotes |
|---|
Claudia Kappen, Associate Editor
|
References |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResults and DiscussionConclusionsSupplementary MaterialAcknowledgementsReferences |
|---|
Amoutzias GD, Bornberg-Bauer E, Oliver SG, Robertson DL. (2006) Reduction/oxidation-phosphorylation control of DNA binding in the bZIP dimerization network. BMC Genomics 7:107.[CrossRef][Medline]
Amoutzias GD, Robertson DL, Oliver SG, Bornberg-Bauer E. (2004) Convergent evolution of gene networks by single-gene duplications in higher eukaryotes. EMBO Rep 5:274–279.[CrossRef][ISI][Medline]
Bertrand S, Brunet FG, Escriva H, Parmentier G, Laudet V, Robinson-Rechavi M. (2004) Evolutionary genomics of nuclear receptors: from twenty-five ancestral genes to derived endocrine systems. Mol Biol Evol 21:1923–1937.
Blomme T, Vandepoele K, De Bodt S, Simillion C, Maere S, Van de Peer Y. (2006) The gain and loss of genes during 600 million years of vertebrate evolution. Genome Biol 7:R43.[CrossRef][Medline]
Boeckmann B, Bairoch A, Apweiler R, et al. (12 co-authors). (2003) The SWISS-PROT protein knowledgebase and its supplement TrEMBL in 2003. Nucleic Acids Res 31:365–370.
Bornberg-Bauer E, Rivals E, Vingron M. (1998) Computational approaches to identify leucine zippers. Nucleic Acids Res 26:2740–2746.
Brendel V and Karlin S. (1989) Too many leucine zippers? Nature 341:574–575.[CrossRef][Medline]
Brenner SE, Chothia C, Hubbard TJ. (1998) Assessing sequence comparison methods with reliable structurally identified distant evolutionary relationships. Proc Natl Acad Sci USA 95:6073–6078.
Bru C, Courcelle E, Carrere S, Beausse Y, Dalmar S, Kahn D. (2005) The ProDom database of protein domain families: more emphasis on 3D. Nucleic Acids Res 33:D212–D215.
Danchin EG, Gouret P, Pontarotti P. (2006) Eleven ancestral gene families lost in mammals and vertebrates while otherwise universally conserved in animals. BMC Evol Biol 6:5.[CrossRef][Medline]
Davidson EH and Erwin DH. (2006) Gene regulatory networks and the evolution of animal body plans. Science 311:796–800.
Dehal P and Boore JL. (2005) Two rounds of whole genome duplication in the ancestral vertebrate. PLoS Biol 3:e314.[CrossRef][Medline]
Deppmann CD, Alvania RS, Taparowsky EJ. (2006) Cross-species annotation of basic leucine zipper factor interactions: insight into the evolution of closed interaction networks. Mol Biol Evol 23:1480–1492.
Eddy SR. (1996) Hidden Markov models. Curr Opin Struct Biol 6:361–365.[CrossRef][ISI][Medline]
Fassler J, Landsman D, Acharya A, Moll JR, Bonovich M, Vinson C. (2002) B-ZIP proteins encoded by the D. melanogaster genome: evaluation of potential dimerization partners. Genome Res 12:1190–1200.
Fong JH, Keating AE, Singh M. (2004) Predicting specificity in bZIP coiled-coil protein interactions. Genome Biol 5:R11.[CrossRef][Medline]
Giot L, Bader JS, Brouwer C, et al. (49 co-authors). (2003) A protein interaction map of D. melanogaster. Science 302:1727–1736.
Grigoryan G and Keating AE. (2006) Structure-based prediction of bZIP partnering specificity. J Mol Biol 355:1125–1142.[CrossRef][ISI][Medline]
Gronemeyer H, Gustafsson JA, Laudet V. (2004) Principles for modulation of the nuclear receptor superfamily. Nat Rev Drug Discov 3:950–964.[CrossRef][ISI][Medline]
Gu YZ, Hogenesch JB, Bradfield CA. (2000) The PAS superfamily: sensors of environmental and developmental signals. Annu Rev Pharmacol Toxicol 40:519–561.
Guindon S and Gascuel O. (2003) A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. Syst Biol 52:696–704.[CrossRef][ISI][Medline]
Haas NB, Cantwell CA, Johnson PF, Burch JB. (1995) DNA-binding specificity of the PAR basic leucine zipper protein VBP partially overlaps those of the C/EBP and CREB/ATF families and is influenced by domains that flank the core basic region. Mol Cell Biol 15:1923–1932.[Abstract]
Hurst HC. (1994) Transcription factors 1: bZIP proteins. Protein Profile 1:123–168.[ISI][Medline]
Johnson JM, Mason K, Moallemi C, Xi H, Somaroo S, Huang ES. (2003) Protein family annotation in a multiple alignment viewer. Bioinformatics 19:544–545.
Klemm JD, Schreiber SL, Crabtree GR. (1998) Dimerization as a regulatory mechanism in signal transduction. Annu Rev Immunol 16:569–592.[CrossRef][ISI][Medline]
Kusserow A, Pang K, Sturm C, Hrouda M, Lentfer J, Schmidt HA, Technau U, von Haeseler A, Hobmayer B, Martindale MQ, Holstein TW. (2005) Unexpected complexity of the Wnt gene family in a sea anemone. Nature 433:156–160.[CrossRef][Medline]
Levine M and Tjian R. (2003) Transcription regulation and animal diversity. Nature 424:147–151.[CrossRef][Medline]
Luscher B. (2001) Function and regulation of the transcription factors of the Myc/Max/Mad network. Gene 277:1–14.[CrossRef][ISI][Medline]
Massari ME and Murre C. (2000) Helix-loop-helix proteins: regulators of transcription in eucaryotic organisms. Mol Cell Biol 20:429–440.
Matys V, Fricke E, Geffers R, et al. (21 co-authors). (2003) TRANSFAC: transcriptional regulation, from patterns to profiles. Nucleic Acids Res 31:374–378.
Miyata T and Suga H. (2001) Divergence pattern of animal gene families and relationship with the Cambrian explosion. Bioessays 23:1018–1027.[CrossRef][ISI][Medline]
Moll JR, Acharya A, Gal J, Mir AA, Vinson C. (2002) Magnesium is required for specific DNA binding of the CREB B-ZIP domain. Nucleic Acids Res 30:1240–1246.
Newman JR and Keating AE. (2003) Comprehensive identification of human bZIP interactions with coiled-coil arrays. Science 300:2097–2101.
Notredame C, Higgins DG, Heringa J. (2000) T-Coffee: a novel method for fast and accurate multiple sequence alignment. J Mol Biol 302:205–217.[CrossRef][ISI][Medline]
Ohno S. (1970) Evolution by gene duplication(Springer, Berlin (Germany)).
Oyake T, Itoh K, Motohashi H, Hayashi N, Hoshino H, Nishizawa M, Yamamoto M, Igarashi K. (1996) Bach proteins belong to a novel family of BTB-basic leucine zipper transcription factors that interact with MafK and regulate transcription through the NF-E2 site. Mol Cell Biol 16:6083–6095.[Abstract]
Popovici C, Roubin R, Coulier F, Birnbaum D. (2005) An evolutionary history of the FGF superfamily. Bioessays 27:849–857.[CrossRef][ISI][Medline]
Ranea JA, Grant A, Thornton JM, Orengo CA. (2005) Microeconomic principles explain an optimal genome size in bacteria. Trends Genet 21:21–25.[CrossRef][ISI][Medline]
Robinson-Rechavi M, Escriva Garcia H, Laudet V. (2003) The nuclear receptor superfamily. J Cell Sci 116:585–586.
Shiu SH and Li WH. (2004) Origins, lineage-specific expansions, and multiple losses of tyrosine kinases in eukaryotes. Mol Biol Evol 21:828–840.
Smolen P, Baxter DA, Byrne JH. (2000) Mathematical modeling of gene networks. Neuron 26:567–580.[CrossRef][ISI][Medline]
van Nimwegen E. (2003) Scaling laws in the functional content of genomes. Trends Genet 19:479–484.[CrossRef][ISI][Medline]
Vinson C, Acharya A, Taparowsky EJ. (2006) Deciphering B-ZIP transcription factor interactions in vitro and in vivo. Biochim Biophys Acta 1759:4–12.[Medline]
Vinson C, Myakishev M, Acharya A, Mir AA, Moll JR, Bonovich M. (2002) Classification of human B-ZIP proteins based on dimerization properties. Mol Cell Biol 22:6321–6335.
Wagner EF. (2001) AP-1—introductory remarks. Oncogene 20:2334–2335.[CrossRef][ISI][Medline]
Wolfe KH. (2001) Yesterday's polyploids and the mystery of diploidization. Nat Rev Genet 2:333–341.[CrossRef][ISI][Medline]
Yamada L, Kobayashi K, Degnan B, Satoh N, Satou Y. (2003) A genomewide survey of developmentally relevant genes in Ciona intestinalis. IV. Genes for HMG transcriptional regulators, bZip and GATA/Gli/Zic/Snail. Dev Genes Evol 213:245–253.[CrossRef][ISI][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  A. M. Reitzel, J. C. Sullivan, N. Traylor-knowles, and J. R. Finnerty Genomic Survey of Candidate Stress-Response Genes in the Estuarine Anemone Nematostella vectensis Biol. Bull., June 1, 2008; 214(3): 233 - 254. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. W. Pinney, G. D. Amoutzias, M. Rattray, and D. L. Robertson Reconstruction of ancestral protein interaction networks for the bZIP transcription factors PNAS, December 18, 2007; 104(51): 20449 - 20453. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||